Design and Synthesis of a Series of Truncated Neplanocin Fleximers

In an effort to study the effects of flexibility on enzyme recognition and activity, we have developed several different series of flexible nucleoside analogues in which the purine base is split into its respective imidazole and pyrimidine components. The focus of this particular study was to synthesize the truncated neplanocin A fleximers to investigate their potential anti-protozoan activities by inhibition of S-adenosylhomocysteine hydrolase (SAHase). The three fleximers tested displayed poor anti-trypanocidal activities, with EC50 values around 200 μM. Further studies of the corresponding ribose fleximers, most closely related to the natural nucleoside substrates, revealed low affinity for the known T. brucei nucleoside transporters P1 and P2, which may be the reason for the lack of trypanocidal activity observed.


Introduction
Modified nucleosides, in particular carbocyclic nucleosides, are potent inhibitors of S-adenosyl homocysteine hydrolase (SAHase) [1]. SAHase is a critical enzyme that hydrolyzes S-adenosyl homocysteine, the byproduct of biomethylations that utilize S-adenosylmethionine (SAM) [2,3]. By inhibiting SAHase, an excess of SAH is produced, which in turn exhibits potent inhibitory effects on methyltransferases [4]. Thus, inhibition of SAHase leads to incomplete methylation of nucleic acids, phospholipids, proteins, and other small molecules, disrupting various biochemical pathways [5]. As a result, carbocyclic nucleosides have proven useful in a number of chemotherapeutic applications [6][7][8].
Neplanocin A (NpcA, Figure 1) and aristeromycin (Ari) are both naturally occurring carbocyclic adenosine analogues that have shown significant antiviral, antiparasitic and anticancer properties [4,7,9,10]. Unfortunately, NpcA and Ari both exhibit deleterious cytotoxicity due to intracellular conversion to their triphosphate forms by adenosine kinase as well as their recognition and metabolism by adenosine deaminase [11][12][13]. Removal of the 4'-CH2OH from Ari and NpcA, as shown in the truncated analogues shown in Figure 1 (R = H), significantly lowers the cytotoxicity [14]. Interestingly, nucleosides with base modifications such as 3-deazaadenosine have also been found to act as substrates, with similar Km's found for adenosine and 3-deazaadenosine [7,10,15]. To date, the truncated 3-deaza analogues of Ari and NpcA (truncated DZNepA, Figure 1) lacking the 4'-hydroxymethyl group have both exhibited greater levels of inhibition than their parent counterparts [5,6,13,16].
More importantly, these compounds have also shown potent inhibition against chloroquine-resistant and chloroquine-susceptible strains of P. falciparum [5]. In protozoan parasites, methylation of the four nucleosides present in the "cap-four" terminal end of mRNA requires SAM as the methyl donor. This cap structure is important for RNA recognition and stability, is highly conserved across almost all protozoan species, and is critical for replication [17][18][19]. Thus, inhibition of SAHase results in an accumulation of SAH, causing methylations to cease, which then disrupts the methylation of the cap structure, thereby providing an important target for the development of potential antiparasitic chemotherapeutics [5].
The Seley-Radtke group has long been interested studying the effects of flexibility on the nucleobase. This flexibility is achieved by "splitting" the purine base into its respective imidazole and pyrimidine (or pyridine) components, which remain connected by a single carbon-carbon bond between the two heteroaromatic moieties. This connectivity allows for free rotation, while still retaining the elements essential for base pairing and molecular recognition [20][21][22][23][24][25][26][27][28]. This modification has led to enhanced enzyme binding and recognition, as well as the ability to overcome point mutations in enzyme binding sites [29,30]. These analogues have also been studied for their potential therapeutic properties [20][21][22][23][24][25][26][27][28]. Interestingly, when the fleximer analogues of adenosine (Flex-A), inosine (Flex-I) and guanosine (Flex-G) were studied in SAHase, which is a flexible enzyme, Flex-A and Flex-I acted as substrates, whereas Flex-G proved to be an inhibitor [22]. This is significant because it is, to our knowledge, the only report of a G-nucleoside inhibiting an adenosine metabolizing enzyme. It has been postulated that this is due to an intramolecular hydrogen bond between the pyrimidine and the 5'-OH of the sugar, which then positions the amino group into the binding site where the amino group on adenosine would normally reside, thus essentially creating an adenosine mimic [22].
Historically, a number of nucleoside analogues have been evaluated for trypanocidal activity [31][32][33][34]. For example, Cai et al. showed that the antiviral drug ribavirin was an inhibitor of Trypanosoma cruzi SAHase [33]. Additionally, 7-deaza-5'-noraristeromycin was shown to be a potent inhibitor of four strains of Trypanosoma brucei [34]. To further explore the potential of base flexibility and antiparasitic activity, we combined the fleximer base with the carbocyclic nucleoside scaffold, to determine whether the flexible base motif would enhance the biological results previously observed with carbocyclic analogues such as NpcA and Ari. Thus, a series of 3-deaza fleximers (compounds 1-3, Figure 1) were designed and synthesized to evaluate their anti-parasitic properties.

Chemistry
As shown in Scheme 1, cyclopentenol 5 was available from known literature procedures starting from D-cyclopentenone 4 [35], which can be obtained following stereospecific reduction to the "down" hydroxyl using Luche reduction conditions [36]. Alcohol 5 was then coupled to 4,5-diiodoimidazole [29] using standard Mitsunobu [37] conditions to give 6. Initially the Mitsunobu reaction was attempted with diisopropylazodicarboxylate (DIAD) and triphenylphosphine (TPP) in dichloromethane at room temperature to yield 5, however only in a 12% yield. Attempts at heating the reaction only served to give additional side products, as well as to lower the yield even further. Changing the solvent to THF increased the solubility of the diiodoimidazole and subsequently resulted in an improved yield of 40%. Unfortunately, contaminates from the byproduct, triphenylphospine oxide (TPPO), still proved to be problematic during purification. Altering the phosphine reagent to DPPE (1,2-bis(diphenylphosphino)ethane) drastically improved the ease in purification. Other coupling methods were also tried, such as using Hendrickson's "POP" reagent, bis(triphenyl)oxodiphosphonium trifluoromethanesulfonate [38], or using bases such as NaH or K2CO3 [39] to form the imidazole nucleophile, proved unsuccessful when compared to the Mitsunobu coupling using DPPE. Next, as shown below in Scheme 2, removal of the 5-iodo group of 6 to give compound 7 was achieved via selective deiodination using ethyl magnesium bromide (EtMgBr) followed by quenching with water. Coupling to the pyridine ring was then accomplished using Stille [40] coupling. The 3-tributyltin-2-chloropyridine was prepared from the commercially available 3-bromo-2chloropyridine. Stille coupling of 7 with the 3-tributyltin-2-chloropyridine provided 8 in a 23% yield, however when copper (I) bromide was used, the yield improved to 71%. Following Stille coupling, transformation of the chloro group into the exocyclic amine group was necessary. Standard procedures using MeOH/NH3 or converting the chloro to an azide using sodium or lithium azide proved A literature search revealed a palladium-assisted method developed by Hartwig using sodium t-butoxide in ammonia saturated 1,4-dioxane [41]. Unfortunately this method also proved unsuccessful. Related to this latter route, Buchwald developed a similar method, where the catalyst is made in situ using a more common phosphine ligand [42]. This method seemed promising since one of the examples utilized 2-chloropyridine, which was successfully converted in a 96% yield [28], but it too proved to be unsuccessful. Use of NaNH2 in ammonia was also tried but the conditions proved to be too harsh and decomposition ensued [43]. Another approach involved converting the chloro group using hydrazine followed by reduction. Initial attempts at reducing the hydrazine employed zinc in acetic acid, but this resulted in a complex mixture that could not be purified. Using titanium chloride (TiCl3) [44] proved to be successful, although there was evidence of some isopropylidene deprotected product(s) as well as protected products, thus treatment of the mixture with dilute TFA in THF gave the desired final product 1.
Next, deaminated compound 10 was obtained from 8 using concentrated acetic acid at high temperature. Although this conversion also led to partial deprotection of the isopropylidene on the 2'-and 3'-hydroxyls, the protected pyridine 10 was the major product. Subsequent deprotection of the isopropylidene of 10 led to the fleximer inosine 3.

Trypanosomiasis Screening
The three NpcA fleximers (1-3) were tested for trypanocidal activity against the laboratory Trypanosoma brucei brucei strain Lister 427, using a standard protocol based on the fluorescent format of 23 doubling dilutions, starting at 500 μM, in 96-well plates. All three fleximers tested displayed very similar activities against this strain, with EC50 values around 200 μM; in contrast, the control drug pentamidine displayed activity in the low nM range (Table 1), consistent with previous results [45,46]. We considered that the relatively low activity might be related to a lack of recognition of these molecules by the T. brucei nucleoside transporters. We therefore investigated whether fleximers in general display reduced uptake kinetics in these parasites, compared to their fixed-ring counterparts ( Figure 2). Using the fleximers [21] most closely related to the original nucleoside substrates, it is clear from Table 2 that fleximers indeed show low affinity for the known T. brucei nucleoside transporters P1 and P2 [47].   Koning [47] and included here for comparison; δ(ΔG 0 ) is the difference in Gibbs free energy of interaction of the nucleoside and the fleximer with the transporter, given in kJ/mol. Transport, mediated by the P1 nucleoside transporter, was measured in the presence or absence of various concentrations of nucleosides (filled symbols) or their corresponding fleximers (open symbols), in the presence of 100 µM adenine to block potential adenosine transport through the P2 transporter. Data shown are the average and SEM of triplicate determinations in a single experiment, representative of three independent experiments with essentially identical outcomes.
It is thus clear that the fleximers generally display about two orders of magnitude less affinity for the T. brucei nucleoside transporters than the corresponding nucleosides, limiting cellular uptake as there are no other nucleoside uptake mechanisms in these parasites than the P1 and P2 systems, although P1 consists of a cluster of multiple genes with slightly divergent sequences [48,49]. In addition, the truncated NpcA fleximers lack a 4'-hydroxymethyl group and an equivalent of the purine N3 residue, and both required for high affinity for P1 [50]. Moreover, the P2 transporter does not recognize any oxopurine nucleoside analogues [47]. The loss of approximately 10 kJ/mol in Gibbs free energy for the fleximer-transporter interaction may in part be due to the increased entropy in the orientation of the fleximer orientation in the binding pocket, as well as the slightly larger volume of the base. We thus conclude that the low effectiveness of the Npc fleximers is at least partially due to unfavorable interactions with the parasite's nucleoside transporters. As important differences exist between nucleoside transporters of even closely related pathogenic parasites including Trypanosoma congolense [51] and Leishmania species [48], it would be worthwhile to follow this study with a wider screening of anti-parasite activity for a diverse panel of protozoa.

General Information
All chemicals were obtained from commercial sources and used without further purification unless otherwise noted. Anhydrous DMF, MeOH, DMSO and toluene were purchased from Fisher Scientific

Anti-Trypanosome Activity
In vitro activity against Trypanosoma brucei was determined using the Alamar blue (resazurin) assay for cell viability exactly as described [52]. Briefly, serial dilutions of test compounds were made in 96-well plates by serial passage of 100 μL of test compound (usually at 2 mM) to 100 μL of HMI9 medium containing 10% fetal bovine serum (Invitrogen), using 2 rows, with the negative control values obtained from wells with 100 μL of medium without test compound. Serial dilutions with pentamidine isethionate (Sigma) were used as positive control. To each well, 100 μL of medium, containing 10 4 culture-adapted bloodstream T. b. brucei (strain Lister 427), was added and the plates were incubated at 37 °C for 48 h after which 20 μL 5 mM resazurin solution was added. Following a further incubation of 24 h at 37 °C, fluorescence was determined in a FLUOstar OPTIMA (BMG Labtech, Aylesbury, UK) fluorimeter with excitation and emission filters at 544 nm and 620 nm, respectively. EC50 values (the effective concentration reducing specific fluorescence by 50%) were calculated by nonlinear regression using the Prism 5 software package (GraphPad, La Jolla, CA, USA).

Transport Assays
Transport assays with bloodstream forms of T. b. brucei were performed exactly as described previously [53,54]. Briefly, transport was initiated by the addition of 100 µL of T. b. brucei bloodstream forms (10 7 cells/mL in assay buffer [53]) to 100 µL of [2,8,5'-3 H]-adenosine (PerkinElmer, Waltham, MA, USA; 54.4 Ci/mmol) pre-mixed with up to 1 mM of test inhibitor in assay buffer. After exactly 10 s the mixture was centrifuged through an oil layer in a microfuge (13,000× g) and the microfuge tubes were flash-frozen in liquid nitrogen. Pellets were cut off and collected in scintillation tubes; after solubilisation in 2% SDS and addition of scintillation fluid, radioactivity was determined in a liquid scintillation counter. Inhibition data were fitted to a sigmoidal curve with variable slope (GraphPad Prism 5.0), allowing for the determination of EC50 values, from which inhibition constants (Ki) were calculated using the Cheng-Prusoff equation, and Gibbs Free Energy using ΔG 0 = −RTln (Ki), as described [52].

Conclusions
The strategy of the work presented herein was to potentially synthesize new and more potent inhibitors of SAHases, thereby disrupting mRNA capping in protozoa as a strategy towards new antiparasitic therapeutics. To this end, characteristics of known SAHase inhibitors such as neplanocin and Aristeromycin were combined, and the nucleoside analogue was given enhanced flexibility using the "fleximer" approach, and added specificity by omitting the N3 equivalent nitrogen residue in the pyrimidine half of the fleximer base group. In addition, the 4'-CH2OH moiety was omitted to reduce general cytotoxicity [10,14]. The data, however, show that the resulting 3-deazaneplanocin fleximers (1-3) displayed only moderate activity in a standardized anti-protozoal test, against Trypanosoma brucei, despite the possibility of this species being vulnerable to inhibition of SAHase [55].
We have previously shown that the trypanocidal action of nucleoside and nucleobase analogues is either enabled or limited by their rate of uptake by specific transport proteins [52][53][54]56,57], and therefore investigated the effect of the fleximer modification on nucleoside transport. We found that the introduction of this modification of the purine ring reduces affinity, and thus presumably translocation rates, for both of the transport systems expressed in bloodstream T. brucei, and conclude that the lack of suitable transporters for these molecules causes (or at least contributes to) the observed lack of trypanocidal potency. However, we have also shown that purine transporters in other protozoan parasites, e.g., Toxoplasma gondii [58], Plasmodium falciparum [59], Leishmania donovani [48], and Trichomonas vaginalis (Natto and De Koning, unpublished data) all have very different substrate-specificity characteristics. Further studies with additional parasites, and the optimization of the inhibitors for enhanced uptake by the parasites, are in progress.